Site:JNR device
From PESWiki
This page is out of date. Please see the AEM website for an update and details -->
http://aemcomp.googlepages.com
Spice simulations clearly show that a diode array or even a single diode rectify Johnson noise. Consider the following circuit, called a JB2.
The following screen shot taken from LTspice is from the above single JB2 circuit using a modified HSMS-2850 diode. The modification is that has 0.45 fF as Cjo. Note that Verginia Diodes Inc. use to make a micro THz diode that was 0.45 fF. Resistor R2 is the output, which as you can see in the following graph is over 1.95 pW DC. Further Spice simulations shows the above JB2 circuit is scalable in that 100 JB2's produces 100 times the DC power. The 1.95 pW may not sound like much power, but 500 billion of such microscopic JB2 bridges would generate ~ 1 watt of DC power. Note that Johnson noise is unchanging relative to size, as P = K T B, where P is power, K is Boltzmann constant, B is bandwidth.
Some debunker's claimed the above power figures are too high due to resistor capacitance. Resistor capacitance decrease with a decrease in resistor size. Furthermore, ten resistors of value 1/10 R in series has approximately 1/10th the net capacitance. Also there are techniques to appreciably prevent the skin effect. High frequency resistors should be long relative to width, and small. Note that present technology goes far above 100 GHz using small resistors. Creating microscopic THz resistors is within the means of present technology.
Note that each JB2 circuit would not conduct that much power or current. Rather, when considering 500 billion microscopic JB2's, such power is spread out, as each diode would conduct a rather small amount of current. Most of such JB2 circuits would be in parallel to generate 5 volts DC since according to Spice each JB2 generators 0.375 mV. So a 500 billion JB2 chip would have ~14,000 in series and ~36,000,000 in parallel to generate ~5.25 volts DC. At 1 watt DC out and 5.25 volts comes to 0.19 amps DC out. At 36,000,000 in parallel comes to 0.19 A / 36,000,000 = 5.3 nA. Each diode would conduct ~ 5.3 nA (1 nA = 1E-9 amps). Furthermore, rectifying Johnson noise would drop the diodes temperature since the net result is energy is being removed, not added.
The above JB2 circuit, less the R2 load, produces 0.375 mV DC from per bridge, or 0.200 mV DC with the R2 load. Note that load resistor does not decrease the JB2 DC output voltage itself, but drops the output voltage. Spice simulations show the JB2 circuit is scalable in that 100 JB2's in series produces 100 times the DC voltage output. Therefore, voltage is not a problem. Fourteen thousand of such JB2 bridges would produce over 5 volts DC. The following screen shot was taken of a LTspice simulation showing the JB2 DC voltage.
The following circuit is four JB2's using unmodified/stock HSMS-2850 diodes.
According to spice simulations JB2 produces ~0.85uV per bridge using HSMS-2850 diodes. The above circuit consists of 4 JB2's, and thus produces 3.4uV. Each JB2 bridge requires just two diodes. Therefore, according to Spice, 1024 diodes should produce 1024/2 * 0.85uV = 0.44mV.
The following spice results of four JB2 bridges charging eight 20pF capacitors for a total 160pF to 3.4uV. The time scale of the entire graph is from 0us to 18us. The voltage scale of the entire graph is from 0uV to 3.6uV.
The following circuit is one JB1 circuit.
The above circuit rectifies Johnson noise caused by R1. The above circuit charges the capacitor to 72nV DC.
The following graph is from 15 of the above full wave bridge rectifiers in series. The capacitor charged to 1.084 uV DC.
The following graph is from 80 full wave bridge rectifiers in series. The capacitor charged to 5.79 uV DC.
The DC voltage rectification appears to be linear relative to the number of JB's in series. Eighty thousand JB's in series should produce 5.79mV.
The following circuit shows how to connect the full wave bridge rectifiers in series.
